Jump to content
  • Nvidia seemingly learns its lesson as RTX 5000 allegedly has 512-bit memory bus (32GB VRAM)


    Karlston

    • 412 views
    • 3 minutes
     Share


    • 412 views
    • 3 minutes

    The current generation of Nvidia graphics cards, the Ada Lovelace (RTX 40-series), are quite interesting in the way the tech media has received it. On one hand, while the higher-end models like the RTX 4090 and the 4080 are generally recommended as great purchases for 4K gaming, the mid-range options consisting of the RTX 4060 Ti and 4060, and especially the former, are heavily criticized due to the low memory capacity and low memory bandwidth they offer, and Nvidia rightfully deserves all of that.

     

    The 4060 Ti was bashed a lot for its 8GB VRAM and 128-bit bus. While 8 gigs of memory buffer at a price point of $400 in itself is already so bad, the chips were running at 18Gbps, and across its 128-bit wide bus, the total memory bandwidth is just 288GB/s. This was more than 35% lower than the 3060 Ti's 448GB/s which was 256-bit wide and the memory chips were 14Gbps. And sadly Ada Lovelace's large pool of L2 cache (32MB on the 4060 Ti compared to 4MB on the 3060 Ti) was not enough to compensate for this.

     

    When one contrasts this to the RTX 3090 vs the 4090, it is clear why the 4090 was such a big jump in performance while the 4060 Ti is barely any faster than its predecessor. While both feature 24GB of memory, the bandwidth of the 4090 is actually higher by around 7.7% (1,008 GB/s vs 936.2 GB/s) , as both are 384-bit wide. Meanwhile, the L2 cache on the 4090 sees a colossal 12 11 times increase (6MB vs 72MB), compared to a four three times increase on the x60 Ti. Hence there is no memory configuration bottleneck on the 4090 but the same cannot be said for the 4060 Ti.

     

    The comparison between the 3090 Ti and the apparently-now-canceled RTX 4090 Ti would have been similar in line with the x90 cards as the memory system is alike. Moving forward, according to leaker kopite7kimi, Nvidia is planning to increase the bus width on its next-gen RTX 50-series cards (allegedly codenamed "Blackwell").

     

    While a narrow bus helps to lower the power draw, Nvidia may feel it needs more bandwidth for its next-gen cards as it is purportedly increasing it from 384-bit to 512-bit. What this means is that the next Nvidia flagship GPU (RTX 5090 or 5090 Ti) will likely have 32GB of VRAM.

     

    With this increased bus interface and capacity, it is hoped that some of the benefits will trickle down to the RTX 5060 and 5060 Ti class of cards as well and Nvidia does not heavily cripple the memory subsystems of these GPUs like on the present 4060 Ti.

     

    Source


    User Feedback

    Recommended Comments

    There are no comments to display.



    Join the conversation

    You can post now and register later. If you have an account, sign in now to post with your account.
    Note: Your post will require moderator approval before it will be visible.

    Guest
    Add a comment...

    ×   Pasted as rich text.   Paste as plain text instead

      Only 75 emoji are allowed.

    ×   Your link has been automatically embedded.   Display as a link instead

    ×   Your previous content has been restored.   Clear editor

    ×   You cannot paste images directly. Upload or insert images from URL.


  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...